Overview

Gait recognition studies how to identify people from the way they walk, even when faces are unavailable or capture conditions vary. Methods learn identity cues from silhouettes, video, body poses or skeletons, and temporal motion; newer work also combines cameras with sensors such as LiDAR and event cameras, or uses language-guided features. Benchmarks such as CASIA-B, SUSTech1K, GREW, and Gait3D test performance across viewpoints, clothing, carrying conditions, and real-world scenes. A central challenge is learning identity features that remain reliable across modalities, sessions, occlusion, and appearance changes without exposing sensitive personal information.

Latest papers

CardsList

32 papers3 papers in the last four weeks, down 40% on the four weeks before. 0.0% of all new papers.

  1. Evaluating Large Language Models for Gait Classification Using Text-Encoded Kinematic Waveforms

    Mar 4, 2026Carlo Dindorf, Jonas Dully, Rebecca Keilhauer +2LLM EvaluationGait Analysis

  2. Sketch It Out: Exploring Label-Free Structural Cues for Multimodal Gait Recognition

    Mar 4, 2026Chao Zhang, Zhuang Zheng, Ruixin Li +1Gait RecognitionMultimodal Fusion